GaN for High-Power RF

UMS, which is jointly owned by Thales and EADS, is moving forward on GaN for some key applications.

United Monolithic Semiconductors (UMS) is a leader in the design, manufacturing and marketing of advanced semiconductor technology for a number of specialized application areas. The company sees advantages in moving from GaAs to GaN for some of the high-power RF applications in which we are active, and is approaching the process qualification phase.

In particular, GaN has the potential to be a contender in high power products between 2 and 20GHz. More specifically:

• For commercial applications, such as power amplifiers (PA) for base stations, which is the largest potential market today, and for new standards appearing at higher frequencies like WIMAX.
• For military applications, GaN PAs are good candidates for radar and counter measures.
• GaN is also attractive for LNA (low noise amplifiers) due to its excellent electrical robustness (high maximal field), which should simplify the system integration and improve overall performance.

Figure 1. Cross sectional TEM (substrate-buffer interface) (Courtesy of Picogiga) Figure 2. Top region (Schottky AlGaN layer)

GaN Advantages

The reasons why GaN is attractive are well known:

• large band-gap
• high maximum field
• high breakdown voltage
• high carrier density and mobility (through heterostructures like AlGaN/GaN to create two-dimensional electron gas)
• high thermal conductivity (in particular when grown on SiC substrates)

GaN Challenges

Currently, the major challenges to more widespread GaN adoption are reliability and price. While reliability is of course paramount, nobody will accept higher priced devices, regardless of the performance improvement that GaN might bring. Furthermore, the prospect of brand-new devices make some customers wary.

The performance already achieved on GaN devices is quite impressive, but in order to make it to the market place, in particular to replace existing solutions like LDMOS in the base-station business, GaN must offer a major leap in performance.

R&D efforts have shown that output power, power density and RF gain performance targets can be achieved. However, PAE (power added efficiency) has proven to be trickier than expected, in particular in the 10 to 20-GHz band.

Substrate Considerations

As mentioned above, price is a primary key to success. Unfortunately, the best performance is achieved today on the most expensive substrates (SiC). Cheaper substrate solutions like silicon exist, but at a clear cost to performance.

Although there certainly is a market at both ends of the scale, an intermediate solution offering a significant price reduction at a minimum performance cost would help capitalize on the big potential of GaN devices. This is where engineered substrates like SopSiC (Silicon on polySilicon Carbide) have an edge. They should alleviate the performance draw-backs of silicon without jeopardizing the cost advantage.

Promising Investigations

While our main focus is SiC, we are now investigating the feasibility (both technical and commercial) of engineered substrates. We have had very promising results and are actively working on the demonstration of engineered substrates in the framework of the EU HYPHEN project.